From: STR . Not even ethernet with hardware PTP implementations, admittedly I haven't seen a NIC for sale with it? I've always wondered if something like https://store.uputronics.com/index.php?route=product/product&path=60_70&product_id=92 is worth the price over the likes of http://www.jrmiller.demon.co.uk/projects/ministd/frqstd.htm _______________________________________________ The LeoNTP is a generation later, lower-power, has Ethernet, doesn't require serial, can be powered over USB or Ethernet, likely has a more sensitive GPS receiver, can service many clients for NTP etc. etc. I think they are aimed at slightly different markets. The quoted price is GPB 250 for the LeoNTP and GBP 260 for the Miller, but the Miller may include UK tax and the LeoNTP not. If the Miller price includes UK tax, the comparison is GBP 260 Miller, GBP 300 LeoNTP. Much as I would like to support another radio amateur's efforts, I think the LeoNTP outclasses the Miller. If you only need a programmable, dual-output GPS-locked frequency source, at GBP 150 (including tax?): http://www.leobodnar.com/shop/index.php?main_page=product_info&products_id=234 Cheers, David -- SatSignal Software - Quality software written to your requirements Web: http://www.satsignal.eu Email: david-taylor@blueyonder.co.uk Twitter: @gm8arv

> Most GBit NIC have hardware timestamping these days. Though I don't know if whether the APU NICs do. I found 6 Intel 1G NICs that list PTP (IEEE 1588) in their specs - Intel 82574, 82576, 82580, I340, I350 and the I210 - https://communities.intel.com/community/wired/blog/2011/07/07/ieee-1588-update Of those only 1 appears to do hardware timestamping for all Ethernet traffic(??) not just PTP - the Intel I340 - https://communities.intel.com/thread/87751 It appears the APU2 c4 model (which I have) has the Intel I210 NICs which do PTP, I'll check with Pascal/PC Engines if this is true. As you and others have said the APU2+GPS via serial won't be a time-nuts standard time server, this isn't the time/clock source for adding a PTP switch to serve PTP NICs on the LAN. In time, perhaps :) From http://www.intel.com/content/www/us/en/embedded/products/networking/i210-ethernet-controller-datasheet.html Advanced Features: — Audio-video bridging • IEEE 1588/802.1AS precision time synchronization Time-nuts standards require implementations and physics/electronics knowhow that are way beyond me, I'm happy to lurk, learn and implement what little I can. The lack of older, now inexpensive frequency standards hardware availability locally here is another barrier besides their shipping weight.

[I'm responding to the list in case I need correcting] On Tue, Aug 23, 2016 at 3:03 AM, STR . <strykar@hotmail.com> wrote: > Old net4501's are inexpensive, they do occasionally pop up on ebay for > around $20-40 but as making the TAPR Clock-Block is not something I can do > I never bought a 4501. > You started with NTP and my responses have all been in that context. The Soekris modification is an example of time-nut pursuit of a goal. A side effect was a very good NTP local clock. If you goal is a very good local clock then that's an approach. If you goal is time transfer using NTP over ethernet then it's over-kill. I'm not sure of your goal. Modern replacements for the 4501 have been discussed previously. E.g. the BeagleBone Black can be clocked externally and has co-processors that can do hi-res time stamps independent of the primary CPU.

Certainly could work, I give the author credit for saving one gps module from e-waste. Hopefully the USB drivers for the ublox module will provide a virtual com port, then all you have to worry about is wiring the PPS line. On Tue, Aug 23, 2016 at 2:40 AM, STR . <strykar@hotmail.com> wrote: > Could another way be to cut out the mini-pci interface like > http://galeria.tech-blog.pl/U-blox_PCI-5S_GPS/U-blox_PCI-5S_ > GPS_10.jpg.html though I'm not sure if he's driving it via serial? > > > -----Original Message----- > From: time-nuts [mailto:time-nuts-bounces@febo.com] On Behalf Of Scott > Stobbe > Sent: Monday, August 22, 2016 10:46 PM > To: Discussion of precise time and frequency measurement < > time-nuts@febo.com> > Subject: Re: [time-nuts] How to get PPS from ublox mini-PCI GPS to APU2 > SoC serial port for ntpd > > The mini-PCI Express Card specification includes quite a bit more than > just a single PCI-Express lane, there is also provisions for I2C (smbus), > USB, LEDs, SIM, etc... While without looking a the physical card or > specification I can't be certain, but I highly doubt it implements a > PCI-Express interface. It is surely just using the USB interface over the > express card slot. The GPS card should enumerate over the USB bus without > any issues. In windows, ublox should provide you with USB drivers for their > modules. > > You will have to experiment to see if the RX/TX lines of the GPS module > are active after a POR, or if it defaults to USB, as that is how it was > intended to be used. But the PPS line definitely needs to be connected to > one of modem control lines of a UART. > > On Sun, Aug 21, 2016 at 11:30 AM, STR . <strykar@hotmail.com> wrote: > > > Thank you! > > Your post has made my choices clear, UART-c might be it! > > A previous poster said there'd be no difference in jitter using any of > > the > > 4 COM ports and this was reassuring. > > I will start studying the pinouts to confirm the pin numbers while I > > wait for the null-modem cable to arrive. > > > > Another question, would connecting PPS, Tx/Rx to the GPIO pins on > > UART-c make U-blox's Windows u-Center software see this as attached to > > a physical serial port? > > If not, how would I present the GPS to Windows via serial? > > I'd like to test updating the firmware and it appears to not work via > > a mini-pci to WWAN 3G USB adapter. > > > > Others have reported: > > "After Emeryth published the pinout, I also started to experiment with > > this board. I found out that you can update the firmware using the > > LEA-5 firmware version 6.02 from the u-blox web site. The module also > > works very well with the patched firmware version EXT_G50_602_LEA-5H. > > bdbfccefb9dbd8395dec7adece53c1f9.bin. This version enables the output > > of raw data for use with rtklib for real-time kinematic and precision > > positioning. > > > > You can then use the standard drivers from the u-blox web site for > > Windows. They provide a virtual COM port. Please note that updating > > the firmware just worked using the physical serial port of the > > Mini-PCIe card, it did not work using the USB adapter." > > > > > > > > -----Original Message----- > > From: time-nuts [mailto:time-nuts-bounces@febo.com] On Behalf Of Paul > > Sent: Tuesday, August 16, 2016 2:27 AM > > To: Discussion of precise time and frequency measurement < > > time-nuts@febo.com> > > Subject: Re: [time-nuts] How to get PPS from ublox mini-PCI GPS to > > APU2 SoC serial port for ntpd > > > > On Mon, Aug 15, 2016 at 5:35 AM, STR . <strykar@hotmail.com> wrote: > > > > > Pardon my ignorance, I'm not sure what COM port the PPS is tied to > > > or what you mean. > > > > > > I think there's some confusion. > > > > Normally the PPS input to Linux (I'm not sure about FreeBSD) is tied > > to the DCD pin in a serial port. The PPS code is connected to the > > interrupt from the DCD. It can be tied to another pin (DSR or CTS I > > forget), a parallel port or a GPIO pin (a la RaspberryPi or > > BeagleBone). Anything other than DCD normally requires a specific > > kernel build. The APU2c has an I/O part connected via an LPC bus. > > It's essentially 4 UARTS. UART-a is connected to a DE-9 connector via > > a level shifter. All signals are present. UART-b is brought to J3 > > unshifted. Even though J3 has five pins only transmit, receive and > > ground are connected. UARTs "c" and "d" are brought to J17 either as > > 18 GPIO signals or 2x8 RS-232 signals (the latter requires > > non-standard bios code to set up the chip). So if we imagine that you > > want to use the DCD pin on UART-c you'd configure the GPIO pins as > > serial and connect to pin 9 on J17. Be advised that the specifics in > the previous may be wrong so check the schematic. > > > > Now if you want to read the correct time as a sentence from the GPS > > you'd connect the Tx/Rx pins on your module to the corresponding pins > > on a 3V3 serial port. Continuing to use UART-c that would be pins 7 and > 8 on J17. > > > > Now regarding jitter. Pascal suggests that the jitter involved in > > using his take on the LPC connected super i/o part might be too high. > > As noted someone said that was the case with the APU1. While I > > wouldn't be surprised if that were still true with the APU2 you might > > find the time "good enough". Trust but verify. > > > > Finally, these boxes are intended to be routers (hence the three > > network > > interfaces) not time-servers and unless you're irrevocably wedded to > > the miniPCIe in APU2 route there are probably better choices for time > servers. > > _______________________________________________ > > time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to > > https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts > > and follow the instructions there. > > _______________________________________________ > > time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to > > https://www.febo.com/cgi-bin/ mailman/listinfo/time-nuts and follow > > the instructions there. > > > _______________________________________________ > time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to > https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts > and follow the instructions there. > _______________________________________________ > time-nuts mailing list -- time-nuts@febo.com > To unsubscribe, go to https://www.febo.com/cgi-bin/ > mailman/listinfo/time-nuts > and follow the instructions there. >

> [I'm responding to the list in case I need correcting] >You started with NTP and my responses have all been in that context. The Soekris modification is an example of time-nut pursuit of a goal. A side effect was a very good NTP local clock. If you goal is a very good local clock then that's an approach. If you goal is time transfer using NTP over ethernet then it's >over-kill. I'm not sure of your goal. >Modern replacements for the 4501 have been discussed previously. E.g. the BeagleBone Black can be clocked externally and has co-processors that can do hi-res time stamps independent of the primary CPU. Not at all Paul, my original question was about the APU2, I plan to use it as a router and I thought the GPS would be a fun addition. The discussion sort of veered towards PTP and I apologize, I should've changed the subject to match.